Tier 3 data centre: what Uptime’s Tier III rating requires and what it means for your hosting

What a Tier 3 data centre is

A Tier 3 data centre, written Tier III in Uptime Institute’s Tier Classification System, is a facility whose power and cooling can be maintained without shutting down the IT equipment it serves. Uptime’s definition reads: “A Tier III data center is concurrently maintainable with redundant components as a key differentiator, with redundant distribution paths to serve the critical environment.” Uptime’s journal describes concurrent maintainability as “generally, the ability to remove any capacity or distribution component from service on a planned basis without impacting IT”.

For your hosting, a Tier III site should not need your servers switched off when a UPS module, a generator, a chiller or a distribution path is serviced. An unplanned failure can still reach your racks, above all while one path is out of service for maintenance; stopping such failures short of IT is what Uptime adds at Tier IV.

Uptime describes itself as “the only licensed firm that can provide these certifications”. A site called Tier III or Tier 3 may hold an Uptime award, or the label may stand for another scheme or the operator’s own description. TIA-942, the Telecommunications Industry Association’s standard, rates sites from Rated-1 to Rated-4; Uptime’s journal says the two organisations agreed in 2014 to separate their systems and that a reference to a TIA rating “may not include the word Tier”.

Tier I to Tier IV: what each level requires

The four Tiers are progressive, and in Uptime’s words “each Tier incorporates the requirements of the lower Tiers.”

Uptime Institute, Tier Classification System page (read 6 October 2026), and Uptime’s journal articles of September 2014 (April 2021 update) and July 2017. The tenant column is our summary.

Component counts do not define a Tier. Datasheets give redundancy as N+1 or 2N, where N is, in Uptime’s words, “the number of components that are minimally required to meet the load demand”. Its journal says Tier Certification “is not a checklist or cookbook” and that Tier IV can be achieved “with just N+1 components, depending on how they are configured and connected to redundant distribution pathways.”

The public grid does not count towards a Tier either. Uptime’s journal states that under the Tier Standard: Topology “the only reliable source of power for a data center is the engine-generator plant”, and that the number of utility feeds “neither predicts nor influences Tier level”. In June 2014, an article in the same journal said that the standard requires on-site fuel for 12 hours of runtime at N load for all Tiers, with the site still meeting its stated Tier objective.

Tier 3 vs Tier 4: concurrently maintainable or fault tolerant

Both levels have redundant components and redundant distribution paths. At Tier III, according to Uptime’s journal, “each and every component needed to support the IT processing environment can be shut down and maintained without impacting IT operations.” Tier IV adds fault tolerance: when equipment fails or a distribution path is interrupted, “the effects of the events are stopped short of the IT operations.” For that, a Tier IV site has several independent and physically isolated systems and continuous cooling, and Uptime adds that “All of the IT equipment must have a fault-tolerant power design to be compatible.”

Uptime also notes that while redundant components or distribution paths of a Tier IV site are shut down for maintenance, “the environment may experience a higher risk of disruption if a failure occurs.” It leaves the choice of level to each organisation’s tolerance for risk.

While one power path at a Tier III site is in maintenance, your equipment runs on the other path alone, and a failure there reaches it. At Tier IV, a single equipment failure is designed to stop short of your racks, provided your servers have power supplies on both paths. Neither level covers an event that disables the whole building, which is what a recovery site elsewhere is for, as our comparison of hot, warm and cold DR sites explains.

Uptime Institute Tier Certification: design, facility and operations

Uptime certifies in three stages, each a prerequisite for the next. The design award is, in the words of Uptime’s journal, “provisional verification until the Tier Certification of Constructed Facility”, which according to Uptime “ensures that your facility has been constructed as designed”. The third stage, Operational Sustainability, assesses how the site is run.

Uptime Institute’s Tier Certification pages (read 6 October 2026); Uptime’s journal, July 2017 (design milestone, revocation).

As a tenant, check which award covers your building and since when. Uptime shows the awards it has issued on a map with a list view and says it has issued “over 4,300 awards in 120+ countries” (October 2026). A design award alone does not confirm how the building was constructed.

Availability percentages: what Uptime says about 99.982%

Availability figures per Tier, such as 99.982% for Tier III, are still quoted for data centres. Uptime’s journal says, in an explainer updated in April 2021, that “In 2009, Uptime Institute removed all references to ‘expected downtime per year’ when using the Tier Standard.” It adds that “The current Tier Standard of Topology does not assign availability predictions to Tier levels”, and names as one reason for the change “the understanding that operational behaviors can have a huge impact on site availability regardless of the technical prowess of the design and build.”

By our arithmetic, 99.982% of a year leaves about 95 minutes of downtime. A figure of that kind belongs in the provider’s service level agreement, which should also define what counts as unavailable, how it is measured, which maintenance windows are excluded and what remedy applies.

EN 50600 availability classes, the European standard

EN 50600 is CENELEC’s series of European standards for data centre facilities and infrastructures. The CEN/CENELEC/ETSI Coordination Group on Green Data Centres approved the 11th edition of its report on 5 November 2024. According to the report’s executive summary, parts 2-2, 2-3 and 2-4 each define “four levels (1 to 4)” of design availability, for the power supply and distribution, the environmental control and the telecommunications cabling. EN 50600-1 “defines the overall availability level of a data centre based on the lowest level of three infrastructures”, so a site with class 4 power and class 3 cabling is class 3 overall.

DKE/GK 719, by its own description the German national committee that develops and updates the DIN EN 50600 standards, wrote in March 2019 that EN 50600 deliberately names no availability figures for its four classes, because in the committee’s view such figures lack a statistical basis and tend to create false expectations. The classes rest, the committee adds, on design principles such as multiple paths and redundancy of components.

Uptime Institute pages and journal (2014 to 2021); CEN/CENELEC/ETSI Coordination Group on Green Data Centres, executive summary of report edition 11 (approved 5 November 2024); DKE/GK 719 statement (20 March 2019); TÜV NORD’s EN 50600 page (11 September 2026).

We found no document from Uptime or CENELEC that maps the classes onto the Tiers, so read a class and a Tier as separate statements.

What a Tier III rating does not cover

A Tier rating describes the power and cooling infrastructure of one site. Uptime lists factors its standards do not cover, “such as building codes, regional weather, security and property usage”, and its journal says that “geographical location does not affect a facility’s Tier level”. A Tier III building on a flood plain keeps its Tier level; site location is one of the three categories of Uptime’s separate Operational Sustainability assessment. Where a second site should be is a decision of its own, covered in our article on how far a DR site should be from the primary data centre.

Our EU Cloud can add a DR site in a second data centre when one site is not enough. Tell us which systems would need it and where they run today.

The criteria Uptime names for a design award are mechanical, electrical, structural and site elements, so ask separately how many carriers enter the building and by which routes. Operations are assessed apart from the topology, and Uptime’s journal warns that “Even the most robustly designed and constructed facilities may experience outages without a well-developed comprehensive management and operation program”.

Your own systems sit outside every facility rating. Servers need two power supplies on separate feeds to use two paths; for equipment with an odd number of cords, Uptime’s journal says the Tier Standard accepts rack-mounted transfer switches. Keep each rack’s draw within what the feeds of one path can carry alone, because during maintenance on one path the other carries everything. The Tier also does not set how many kilowatts a rack position can draw and cool, which decides whether GPU servers fit; our guide to power and cooling for a GPU rack covers those checks.

How to check a hosting provider’s Tier claim

Collect the following from each provider before you compare offers.

1. Ask which scheme stands behind the label, whether an Uptime award, a TIA-942 rating, an EN 50600 certificate or the provider’s own description.
2. Check the award in Uptime’s list of issued Tier Certifications, or on the certificate, for the building, the stage, the Tier or class and the date.
3. Find out how planned maintenance on power and cooling is announced, and whether customer equipment has had to be shut down for it.
4. Request the number of power feeds per rack, the maximum load per feed and the rule for single-corded equipment.
5. Confirm the fuel stored on site, the refuelling arrangements and which carriers enter the building by which routes.
6. Read the availability commitment in the service level agreement, with its measurement, exclusions and remedy.

Our EU Cloud runs in Baltneta’s Tier-3 data centres in Lithuania, certified to ISO 27001 and PCI DSS Level 1. Send us your supplier questionnaire, and we will state precisely which entity holds which certificate.

Which hosting model fits your workloads is a separate choice, set out in our comparison of IaaS, private cloud and hybrid set-ups.

What we do

Our EU Cloud hosts guaranteed compute or a private cloud in Baltneta’s Tier-3 data centres in Lithuania, with ISO 27001 and PCI DSS Level 1, 2N power redundancy and diesel generators, and with data and backups kept in the EU. You choose IaaS with guaranteed vCPU, RAM and storage on the VMware vSphere platform, a dedicated private cloud or a hybrid set-up with replication between sites, and when needed a DR site in a second data centre. Our engineering partner Vixen.UNO handles migration and support: systems move step by step, in agreed maintenance windows with a rollback plan, and support runs under an SLA. Our security and compliance page sets out where data is processed and which certificates the data centres hold. The first call is free of charge, and you leave it with 2 to 3 configuration options and an indicative monthly invoice.

FAQ